How to Run GUI Data Mining Software on a Mac
For those of us who must have some form of Windows our preferred solution is to mount a Virtual Machine on the Mac host. I am now writing on my MacBook Pro running Parallels 7 but some others in the company prefer VMware. I am perfectly happy with the Parallels approach and test new versions of our software routinely this way, with no problems or complaints. Using VMs I have my choice of old-fashioned XP (my default), or a newer 64-bit Windows (Windows 7 or Server 2008). Since I must leave a reasonable amount of RAM available for the Mac host I cannot allocate all available RAM to the VM and with only 8 GB of RAM I am sometimes constrained when working with large data sets.
So, for some largest data sets for which I need to access to all of my RAM I don't bother to mount the VM at all and work with a nongui 64-bit native Mac version of our software from a terminal window. This requires creating command scripts but if I save the grove files I can view all the results on a Windows GUI SPM when I have completed my runs.
So, what do we recommend for those wanting to work on your Macs? First, be aware that you certainly do not need to buy a dedicated Windows machine. If you don't have enough RAM for heavy duty analytics consider either upgrading your existing Mac with extra memory or buy a new Mac! Second, choose VMWare or Parallels for working with our GUI software. Either one costs about $79.00. Run a 64-bit version of Windows if you are going to need to work with training data files larger than 1 GB. While the Mac Pro is expensive the idea of 12-core 64 GB RAM machine under your desk has got to be appealing, especially since we have versions of SPM that can leverage both the RAM and multi-thread the cores.
Have fun upgrading!